Relay protection-oriented intelligent electronic transformer self-calibration system

Abstract

The invention belongs to the technical field of power system measurement, and discloses a relay protection-oriented intelligent electronic transformer self-calibration system, which comprises a reference construction module used for analyzing an initial error and a drift coefficient of an electronic transformer at the initial stage of commissioning and generating a digital gene vector; the drifting prediction module is used for performing drifting prediction on the historical operation data and the digital gene vector of the electronic transformer and determining a drifting prediction value and drifting confidence of the electronic transformer in a prediction domain; the game decision module is used for acquiring a calibration proposal and corresponding confidence from the topology check domain and the electrical quantity association domain, performing cross-domain confidence game on each domain, and determining a dynamic weight distribution strategy and a calibration decision; the calibration execution module is used for correcting the sampling value of the electronic transformer based on the calibration decision; and the strategy learning module is used for carrying out long-term income evaluation on the intelligent electronic transformer, carrying out dynamic adjustment on the dynamic weight distribution strategy, and enhancing the sustainable calibration capability.

Description

Technical Field

[0001] This invention relates to the field of power system measurement technology, and more specifically, to a self-calibration system for intelligent electronic instrument transformers for relay protection. Background Technology

[0002] In the construction of digital substations and smart grids, electronic instrument transformers are widely used due to their wide dynamic range, lack of magnetic saturation, and good bandwidth adaptability. To ensure that relay protection devices have good reliability and response accuracy in real fault events, the measurement accuracy of electronic instrument transformers must meet the accuracy criteria requirements of relay protection devices for voltage and current measurement values ​​over a long period of time.

[0003] In existing technologies, most electronic current transformers typically employ online comparison techniques for calibrating measurement accuracy. This involves installing a high-precision standard current transformer as a benchmark and continuously monitoring the output data during the operation of the electronic current transformer. When the output deviates from the standard transformer's tolerance by more than a set limit, an alarm or calibration command is triggered. However, this method lacks predictive judgment of the time-varying characteristics of transformer errors and cannot perform proactive calibration before errors exceed limits. For example, under high-temperature conditions, a 220kV line protection electronic current transformer experiences a small but continuous temperature drift error due to changes in the resistance of its internal Rogowski coil winding caused by rising ambient temperature. This error gradually accumulates over time. When this error approaches a set deviation threshold, existing comparison methods can only initiate a passive calibration process after the error actually exceeds the limit. They cannot predict and actively intervene during the stage when the error trend intensifies. This leaves the relay protection device in a state of decreased accuracy for a short period before the calibration response is triggered, posing a risk of misjudgment in the early stages of a fault.

[0004] In view of this, the present invention proposes an intelligent electronic instrument transformer self-calibration system for relay protection to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a self-calibration system for intelligent electronic instrument transformers for relay protection, comprising: The benchmark construction module is used to analyze the initial error and drift coefficient of electronic transformers in the early stage of commissioning and generate digital gene vectors. The drift prediction module is used to predict the drift of the electronic transformer based on its historical operating data and digital gene vector, determine the predicted drift value of the electronic transformer in the prediction domain, and determine the drift confidence level based on the inherent uncertainty in the digital gene vector. The game decision module is used to obtain calibration proposals and corresponding confidence levels from the topology verification domain and the electrical quantity association domain, respectively, to conduct cross-domain confidence game for each domain, and to determine the calibration decision based on the determined dynamic weight allocation strategy. The calibration execution module is used to correct the sampled values ​​of the electronic instrument transformer based on the calibration decision, and output the corrected sampled values ​​to the relay protection device. The strategy learning module is used to evaluate the long-term benefits of smart electronic transformers that have performed calibration decisions, and to dynamically adjust the dynamic weight allocation strategy in cross-domain confidence games based on the obtained evaluation results.

[0006] Furthermore, the specific process for generating digital gene vectors is as follows: Set several operating conditions including specific ambient temperature values ​​and specific primary current load levels; apply a composite test signal to the primary side of the electronic transformer under each operating condition. The composite test signal contains standard parameters at each frequency point; synchronously record the synchronous sampling value sequence of the secondary side of the electronic transformer and perform discrete Fourier transform to obtain the output parameters. The standard parameters at each frequency point are compared with the corresponding output parameters to determine the initial error of each electronic transformer at each frequency point; the parameter fitting is performed on the variation law of the initial error at each frequency point with the operating conditions to determine the temperature drift coefficient, load influence coefficient and the inherent uncertainty of the fitting process. The initial errors, temperature drift coefficients, load influence coefficients, and inherent uncertainties are summarized to generate digital gene vectors for each electronic transformer.

[0007] Furthermore, the specific process for determining the drift prediction value of the electronic transformer in the prediction domain is as follows: Time series analysis is performed on the historical operating data of each electronic instrument transformer to obtain the predicted values ​​of ambient temperature and load current at specific future moments. Based on the temperature drift coefficient and the predicted load current, the predicted ambient temperature and the predicted load current at each frequency point are multiplied to determine the temperature effect drift component and the load effect drift component at each frequency point. The temperature effect drift component and the load effect drift component are superimposed with their corresponding initial errors to determine the drift prediction value of the electronic transformer in the prediction domain. The drift prediction value includes the amplitude error prediction value and the phase error prediction value.

[0008] Furthermore, the specific process for obtaining the predicted values ​​of ambient temperature and load current at a specific future time is as follows: Time series smoothing decomposition was performed on the environmental temperature series and load rate series of historical operating data to determine the horizontal component, trend component and seasonal component of each series at the current moment. The horizontal and trend components of each sequence at the current moment are extrapolated forward to obtain the basic predicted values ​​for each specific future moment; the basic operating condition predicted values ​​are superimposed and adjusted based on the repetitive pattern of the corresponding seasonal components to obtain the predicted values ​​of ambient temperature and load current for each electronic transformer at a specific future moment.

[0009] Furthermore, the specific process for determining the drift confidence level based on the inherent uncertainty in digital gene vectors is as follows: The missing rate and anomaly rate of the ambient temperature series and load rate series in historical operating data are analyzed to determine the basic confidence level of each electronic transformer. Based on the basic confidence level, the inherent uncertainty of the temperature drift coefficient and load influence coefficient are corrected. The uncertainty of the drift prediction value of the corrected temperature drift coefficient and load influence coefficient is quantified to determine the confidence level of the drift prediction value.

[0010] Furthermore, the specific process of obtaining calibration proposals and confidence levels from the topology verification domain and the electrical quantity correlation domain, respectively, is as follows: Acquire the power grid topology data where the electronic instrument transformer is located and the current measurement data of the nodes in the power grid topology; verify the current measurement data using Kirchhoff's current law based on the power grid topology data, and obtain the current imbalance residual of the node; screen out the nodes whose absolute value of the current imbalance residual exceeds the preset residual threshold, and calculate the proportion of the current imbalance residual of each connected electronic instrument transformer in the node. Based on the proportional distribution of current imbalance residuals, the first corrected prediction value of the electronic transformer in each node is obtained. The first corrected prediction value includes the first amplitude correction prediction value and the first phase correction prediction value. These are then summarized to form a calibration proposal for the topology verification domain. The confidence level of the topology verification domain is determined based on the integrity and update frequency of the power grid topology data. The apparent power calculation value of the electronic transformer in each node and the apparent power measurement value of the merging unit are calculated; the relative error between the apparent power measurement value and the apparent power calculation value is calculated, and nodes whose absolute value of the apparent relative error exceeds the preset error threshold are screened out; the second corrected prediction value of the electronic transformer is calculated based on the relative error of each node. The second corrected prediction value includes the second amplitude correction prediction value and the second phase correction prediction value. The two values ​​are summarized to determine the calibration proposal of the electrical quantity correlation domain. The confidence level of the electrical quantity association domain is determined based on the synchronization accuracy of the electronic instrument transformer and the merging unit, as well as the data refresh rate.

[0011] Furthermore, the specific process of conducting cross-domain confidence game across different domains is as follows: Enumerate all possible combinations of the prediction domain, topology verification domain, and electrical quantity association domain, and calculate the weighted average of the calibration proposals in each domain combination according to their corresponding confidence levels to obtain the coalition value of each domain combination. Iterate through all domain combinations containing a specific domain, calculate the expected average increment of the alliance value before and after the specific domain is added to each domain combination, and determine the contribution value of each domain; normalize the contribution value of each domain to determine the dynamic weight allocation strategy; based on the dynamic weight allocation strategy, perform weighted summation on the predicted values ​​of each domain of the electronic transformer to obtain the calibration value of each electronic transformer; summarize the electronic transformers whose absolute calibration value exceeds the preset discrimination threshold and generate an execution decision.

[0012] Furthermore, the specific process of correcting the sampled values ​​of the electronic instrument transformer based on the calibration decision is as follows: The calibration decision is analyzed to determine the electronic transformer to be calibrated and the calibration quantity; the calibration quantity is converted into amplitude calibration coefficient and phase calibration offset, and then sent to the corresponding merging unit. The merging unit performs real-time calibration on each original sampling sequence of the electronic transformer to be calibrated based on the amplitude calibration coefficient and the phase calibration offset, and encapsulates the calibrated sampling values ​​into a sampling value message and outputs it to the relay protection device.

[0013] Furthermore, the specific process for evaluating the long-term benefits of intelligent electronic instrument transformers that have already undergone calibration decisions is as follows: Obtain relay protection action records for all associated calibrated smart electronic instrument transformers during the evaluation period; relay protection action records include correct actions and incorrect actions; obtain the action accuracy rate of the relay protection device based on the relay protection action records; The number of electricity metering disputes caused by out-of-tolerance measurements of electronic instrument transformers during the synchronous evaluation period is counted to determine the dispute-free rate; the geometric mean of the action accuracy rate and the dispute-free rate is calculated to obtain the long-term benefit evaluation value of the dynamic weight allocation strategy; the long-term benefit evaluation value is compared with the preset evaluation threshold, and if the long-term benefit evaluation value is less than the preset threshold for several consecutive evaluation periods, the dynamic weight allocation strategy is updated.

[0014] Furthermore, the specific process for updating the dynamic weight allocation strategy is as follows: Compare the confidence levels of the prediction domain, topology verification domain, and electrical quantity association domain, and identify the domain with the lowest confidence level as the domain to be adjusted; calculate the relative difference between the confidence level of the domain to be adjusted and the average confidence level to determine the confidence contribution deviation of the domain to be adjusted. The confidence contribution deviation is mapped based on the mapping rule to determine the reduction coefficient of the domain to be adjusted. The mapping rule sets the reduction coefficient corresponding to each confidence contribution deviation threshold interval. The reduction factor is loaded into the next round of cross-domain confidence game to reduce the alliance value corresponding to the domain to be adjusted, thereby completing the adjustment of the dynamic weight allocation strategy.

[0015] The technical effects and advantages of the intelligent electronic instrument transformer self-calibration system for relay protection according to the present invention are as follows: 1. This invention, by setting up a benchmark construction module, applies multi-condition composite test signals to electronic instrument transformers during the initial commissioning phase and performs parameter fitting to generate a digital gene vector covering the initial error, temperature drift coefficient, load influence coefficient, and inherent uncertainty at each frequency point. This vector serves as a permanent identifier for the individualized error characteristics of the instrument transformer. Through time-series prediction using historical environmental temperature and load data, and based on the environmental influence factors at future moments, combined with the digital gene vector, quantitative prediction of the future multi-frequency drift error of each instrument transformer is achieved. The drift confidence of the predicted values ​​is evaluated using uncertainty parameters, providing a predictive basis and reference strength for subsequent calibration actions. This invention solves the problem of passive error correction that traditionally requires calibration to be triggered only when measured values ​​exceed limits. It enables early intervention before actual error fluctuations reach the tolerance, effectively ensuring the response accuracy of relay protection devices in the early stages of faults and improving the overall system's initial stability and maintenance efficiency.

[0016] 2. This invention constructs a multi-source calibration proposal game mechanism that integrates the prediction domain, topology verification domain, and electrical quantity association domain. Based on the confidence level of each domain, a dynamic weight allocation strategy is generated, effectively quantifying and integrating the contributions of different calibration paths. By statistically analyzing long-term operational performance indicators such as relay protection operation accuracy and electrical metering disputes, the fusion strategy is periodically evaluated and dynamically optimized. This allows for automatic adjustment of the importance weights of each data domain when operating conditions or main error causes change, ensuring that the final calibration decision always maintains convergence and adaptability. It achieves dynamic collaboration, self-feedback weight adjustment, and sustainable calibration capabilities across multiple decision sources, enhancing the intelligence level and long-term applicability of the electronic transformer calibration system. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of a module of an intelligent electronic instrument transformer self-calibration system for relay protection according to the present invention. Figure 2 This is a flowchart illustrating the process of determining the dynamic weight allocation strategy in this invention. Figure 3 This is a schematic diagram of the calibration process of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: Please refer to Figures 1-3 As shown in the figure, the main design contents of the intelligent electronic transformer self-calibration system for relay protection described in this embodiment are as follows: When electronic instrument transformers operate in a relay protection environment for a long time, their measurement accuracy is easily affected by factors such as ambient temperature, load fluctuations, and device aging, resulting in slow error drift with cumulative and time-varying characteristics.

[0020] To address this type of drift, existing online comparative calibration technologies mainly employ a passively triggered response mechanism after the deviation exceeds the limit. This mechanism cannot effectively identify the accumulation rate and trend of potential errors, resulting in the inability to implement proactive intervention when the error is close to but has not yet exceeded the tolerance. Within short-term sensitive operating windows such as fault initiation and the leading edge of protection actions, the device may be in an uncalibrated drift state, affecting the accuracy of relay protection judgments and thus creating a prediction blind spot for the calibration strategy.

[0021] Based on this, a self-calibration system for intelligent electronic instrument transformers for relay protection is designed, including: The benchmark construction module is used to analyze the initial error and drift coefficient of electronic transformers in the early stage of commissioning and generate digital gene vectors. The specific process of generating digital gene vectors is as follows: Several operating conditions with specific ambient temperature values ​​and primary current load levels are set up. The specific ambient temperature values ​​and primary current load levels are used to simulate the influence of temperature and load on the accuracy of electronic instrument transformers, respectively. Different ambient temperature values ​​and different load points are combined to simulate various working environments that electronic instrument transformers may encounter in actual operation.

[0022] It should be explained that the ambient temperature value should be selected from a typical point covering the planned operating range, and the primary current load level should be selected from a typical grade point of the rated current. For example, the ambient temperature values ​​are set to: -20℃, 0℃, +25℃, +50℃, +70℃; the primary current load level is set to: 30% of rated current, 60% of rated current, 100% of rated current.

[0023] A composite test signal is applied to the primary side of the electronic instrument transformer under each operating condition. The composite test signal contains standard parameters for each frequency point. The composite test signal is synthesized from a standard power frequency fundamental wave and a series of specific integer harmonic components, covering typical frequency points within the operating frequency band of interest of the electronic instrument transformer. The standard parameters for each frequency point in the composite test signal are known standard amplitude and standard phase.

[0024] For example, the frequency points of the composite test signal can be selected as: 50Hz (fundamental wave), 150Hz, 250Hz, 350Hz, and 450Hz.

[0025] The synchronous sampling value sequence of the secondary side of the electronic transformer is recorded synchronously and a discrete Fourier transform is performed to obtain the output parameters. During each composite test signal injection process, the synchronous sampling value sequence of the complete power frequency cycle output by the secondary side of the electronic transformer is continuously acquired. A discrete Fourier transform is performed on each set of synchronous sampling value sequences to extract the output amplitude and output phase at each frequency point to form the output parameters.

[0026] The standard parameters at each frequency point are compared with the corresponding output parameters to determine the initial error of each electronic transformer at each frequency point. The standard parameters (standard amplitude and standard phase) at each frequency point in the composite test signal are used as reference values. The output parameters (output amplitude and output phase) are used as measured values. Then, the difference between the output amplitude and the standard amplitude at each frequency point is calculated to obtain the initial amplitude error at that frequency point. At the same time, the difference between the output phase and the standard phase at each frequency point is calculated to obtain the initial phase error at that frequency point. The initial amplitude errors and initial phase errors of all frequency points are summarized to form the initial error of each frequency point.

[0027] The parameters of the initial error at each frequency point under varying operating conditions were fitted to determine the temperature drift coefficient, load influence coefficient, and inherent uncertainty of the fitting process. The initial errors at each frequency point under all operating conditions were then compiled with the corresponding ambient temperature and primary current load levels to form a dataset containing three dimensions: ambient temperature, primary current load level, and initial error. A least squares method was used to perform a binary linear regression fitting on the dataset, resulting in a fitting plane with ambient temperature and primary current load levels as independent variables and initial error as the dependent variable. The coefficient of the ambient temperature term was extracted from the regression coefficients of this fitting plane and defined as the temperature drift coefficient for that frequency point. The coefficient of the primary current load level term was also extracted and defined as the load influence coefficient for that frequency point. The standard error of the temperature drift coefficient and the standard error of the load influence coefficient were obtained from the least squares fitting results as the inherent uncertainty of the temperature drift coefficient and the load influence coefficient, respectively.

[0028] The initial errors, temperature drift coefficients, load influence coefficients, and inherent uncertainties are summarized to generate a digital gene vector for each electronic transformer. A structured data object is created as the digital gene vector for each electronic transformer. This data object contains the following explicitly identified fields: a list of frequency points (recording all tested frequency points), an initial error mapping (mapping each frequency point to its corresponding initial amplitude error and initial phase error), a temperature drift coefficient mapping (mapping each frequency point to its corresponding temperature drift coefficient), a load influence coefficient mapping (mapping each frequency point to its corresponding load influence coefficient), and an inherent uncertainty field (recording the temperature drift coefficient uncertainty and load influence coefficient uncertainty throughout the fitting process).

[0029] The drift prediction module is used to predict the drift of the electronic transformer based on its historical operating data and digital gene vector, determine the predicted drift value of the electronic transformer in the prediction domain, and determine the drift confidence level based on the inherent uncertainty in the digital gene vector. The specific process for determining the drift prediction value of the electronic instrument transformer in the prediction domain is as follows: Time series analysis is performed on the historical operating data of each electronic instrument transformer to obtain the predicted values ​​of ambient temperature and load current at specific future moments.

[0030] Based on the temperature drift coefficient and the predicted load current, the predicted ambient temperature and the predicted load current at each frequency point are multiplied to determine the temperature effect drift component and the load effect drift component at each frequency point. The temperature drift coefficient and the load influence coefficient of all frequency points are read from the digital gene vector. The temperature drift coefficient of each frequency point is multiplied by the predicted ambient temperature, and the result of each multiplication is the temperature effect drift component at each frequency point. The load influence coefficient of each frequency point is multiplied by the predicted load current, and the result of each multiplication is the load effect drift component at each frequency point.

[0031] For example, the predicted ambient temperature for an electronic instrument transformer is 28°C, and the predicted load current is 75% of the rated current. The temperature drift coefficient (amplitude direction) at the frequency point (50Hz) is +0.001% / ℃, and the load influence coefficient (amplitude direction) is −0.002% / percentage load. Therefore, the temperature effect drift component is 28 × 0.001 = +0.028%, and the load effect drift component is 75 × (−0.002) = −0.15%.

[0032] The temperature effect drift component and the load effect drift component are superimposed with their corresponding initial errors to determine the predicted drift value of the electronic transformer in the prediction domain. The predicted drift value includes the predicted amplitude error value and the predicted phase error value. The initial amplitude error and initial phase error of each frequency point are obtained from the digital gene vector. The temperature effect drift component and the load effect drift component of each frequency point are algebraically added to obtain the comprehensive drift amount of each frequency point. The comprehensive drift amount is algebraically superimposed with the initial amplitude error of the corresponding frequency point to determine the predicted amplitude error value of that frequency point. The comprehensive drift amount is algebraically superimposed with the initial phase error of the corresponding frequency point to determine the predicted phase error value of each frequency point. The predicted amplitude error value and the predicted phase error value of all frequency points are summarized to form the complete predicted drift value of the electronic transformer in the prediction domain.

[0033] For example, for an electronic sensor (50Hz): temperature effect drift component (amplitude: +0.028%), load effect drift component (amplitude: -0.15%); initial amplitude error (-0.03%), initial phase error (+0.2%); total amplitude drift: (+0.028%) + (-0.15%) = -0.122%; After algebraic superposition: Predicted amplitude error: (-0.122A) + (-0.03A) = -0.152%; The predicted phase error is determined as: (-0.122) + (+0.2) = +0.078%. The specific process for obtaining the predicted values ​​of ambient temperature and load current at a specific future time is as follows: The environmental temperature series and load rate series of historical operating data are subjected to time series smoothing decomposition. Each series is split into three components in time index order: a horizontal component representing the current benchmark, a trend component representing long-term changes, and a seasonal component representing periodic fluctuations.

[0034] The horizontal and trend components of each sequence at the current moment are extrapolated forward to obtain the basic predicted values ​​for specific future moments. The number of periods contained in the time interval between the current and future moments is defined. The current ambient temperature trend component is multiplied by the number of periods in the time interval, and then added to the horizontal component of the ambient temperature sequence to obtain the basic predicted value of the ambient temperature. Similarly, the current load rate trend component is multiplied by the number of periods in the time interval, and then added to the horizontal component of the load rate to obtain the basic predicted value of the load rate.

[0035] Based on the repetitive patterns of the corresponding seasonal components, the predicted values ​​of the basic operating conditions are superimposed and adjusted to obtain the predicted values ​​of ambient temperature and load current for each electronic transformer at a specific future time. The component value corresponding to the phase position at a specific future time is extracted from the seasonal components of the ambient temperature sequence, and this component value is added to the basic predicted value of the ambient temperature to obtain the predicted value of the ambient temperature. Simultaneously, the component value corresponding to the phase position at a specific future time is extracted from the seasonal components of the load rate sequence, and this component value is added to the basic predicted value of the load current to obtain the final predicted value of the load current.

[0036] The specific process for determining drift confidence based on the inherent uncertainty in digital gene vectors is as follows: The missing rate and anomaly rate of the ambient temperature series and load rate series in historical operating data are analyzed to determine the basic confidence level of each electronic transformer. The basic confidence level is used to quantify the reliability of historical operating data. The total number of data points and the number of missing data points in the ambient temperature series and load rate series within the analysis period are counted. The proportion of missing data points in each series to the total number of data points is calculated to obtain the corresponding missing rate (ambient temperature missing rate, load missing rate). Data points in the ambient temperature series and load rate series with numerical fluctuations greater than the normal statistical expectation range are identified, and the anomaly rate (ambient temperature anomaly rate, load anomaly rate) is determined by counting the proportion of the identified points to the total number of points.

[0037] The total defect rate of ambient temperature is obtained by adding the ambient temperature missing rate and the ambient temperature abnormal rate. The total defect rate of load rate is obtained by adding the load missing rate and the load rate abnormal rate. The total defect rate of ambient temperature is calculated by arithmetically averaging the total defect rate of load rate. The basic confidence level is determined by using "basic confidence level = 1 - comprehensive defect rate".

[0038] It should be explained that the higher the baseline confidence level, the higher the reliability of historical operating data, the higher the data quality used to predict ambient temperature and load current values, and the more reliable the prediction results.

[0039] The inherent uncertainties of the temperature drift coefficient and the load influence coefficient are corrected based on the baseline confidence level. The baseline confidence level is multiplied by the inherent uncertainties of both the temperature drift coefficient and the load influence coefficient to obtain the corrected uncertainties of the temperature drift coefficient and the load influence coefficient, respectively. The uncertainty of the drift prediction value due to the corrected temperature drift coefficient and load influence coefficient is quantified to determine the confidence level of the drift prediction value. The variance contribution of the temperature component is obtained by multiplying the square of the predicted ambient temperature value by the square of the uncertainty of the corrected temperature drift coefficient; the variance contribution of the load component is obtained by multiplying the square of the predicted load current value by the square of the uncertainty of the corrected load influence coefficient; the variance contributions of the temperature component and the load component are added together and the square root is taken to obtain the standard uncertainty of the drift prediction value. The standard uncertainty is then determined to be an output value between 0 and 1 using a decreasing function, forming the confidence level of the drift prediction value.

[0040] The game decision module is used to obtain calibration proposals and corresponding confidence levels from the topology verification domain and the electrical quantity association domain, respectively, to conduct cross-domain confidence game for each domain, and to determine the calibration decision based on the determined dynamic weight allocation strategy. The specific process for obtaining calibration proposals and confidence levels from the topology verification domain and the electrical quantity correlation domain, respectively, is as follows: The system acquires the power grid topology data where the electronic instrument transformer is located, as well as the current measurement data of the nodes in the power grid topology. The power grid topology describes the connection relationship of various electrical components in the power grid, abstracting electrical connection points such as bus connection points, line endpoints, and transformer winding ends as nodes. The power grid topology data includes the node number, the connection relationship between nodes, and the identification information of the electronic instrument transformers connected to each connection line. The system obtains the current phasor measurement values ​​of each node by uploading the sampling value messages in real time through the merging unit connected to the secondary side of the electronic instrument transformer.

[0041] Based on the power grid topology data, Kirchhoff's current law is used to verify the current measurement data and obtain the current imbalance residuals of the nodes. Current balance verification is performed on each node. Taking each node as a unit, the current vector sum of the lines connected to that node is counted. Kirchhoff's current law is used to determine whether the current at the node satisfies the relationship that "the sum of the inflow currents equals the sum of the outflow currents". Nodes that do not satisfy the current conservation in the statistical results are extracted, and the vector difference between the total inflow and outflow currents of the node is calculated to determine the current imbalance residuals of the extracted nodes.

[0042] Nodes whose absolute value of current imbalance residual exceeds a preset residual threshold are selected, and the proportion of each connected electronic transformer in the node to the current imbalance residual is calculated. Based on the absolute value of the measured current amplitude of each electronic transformer at the node relative to the sum of the absolute values ​​of the measured current amplitude of all connected electronic transformers, the proportion of the electronic transformer to the current imbalance residual is determined.

[0043] It should be noted that electronic current transformers with larger absolute values ​​of measured current amplitudes account for a larger proportion of the total number of such transformers.

[0044] It should be explained that the preset residual threshold is based on the statistical analysis of the node current imbalance residual sequence during historical normal operation; it is used to effectively distinguish between normal random measurement fluctuations and significant anomalies caused by excessive errors in electronic transformers.

[0045] Based on the proportional allocation of current imbalance residuals, the first corrected prediction value of the electronic transformer in each node is obtained. The first corrected prediction value includes the first amplitude correction prediction value and the first phase correction prediction value. These are summarized to form a calibration proposal for the topology verification domain. The current imbalance residuals of the nodes are allocated according to their corresponding proportions. The amplitude of the allocated current imbalance residual vector is divided by the fundamental current amplitude of the electronic transformer itself to obtain the first amplitude correction prediction value. The phase of the allocated current imbalance residual vector is directly used as the first phase correction prediction value.

[0046] For example, the current imbalance residual at node N1 is +5A; (current reference direction convention: positive for current flowing into the node, negative for current flowing out). The measured values ​​of the three branches connected to node N1 and the electronic transformers are as follows: Electronic transformer T1 (incoming line) measured current: +100A; Electronic transformer T2 (outgoing line 1) measured current: -60A; Electronic transformer T3 (outgoing line 2) measured current: -40A; The absolute values ​​of each measured current are: |T1|=100A, |T2|=60A, |T3|=40A; the sum of the absolute values ​​is: 100A+60A+40A=200A. The ratios of the current transformers are: T1 = 100A / 200A = 0.5; T2 = 60A / 200A = 0.3; T3 = 40A / 200A = 0.2; The total unbalanced residual (+5A) of the nodes is distributed proportionally as follows: the first corrected prediction value of electronic transformer T1 = +5A × 0.5 = +2.5A; the first corrected prediction value of electronic transformer T2 = +5A × 0.3 = +1.5A; the first corrected prediction value of electronic transformer T3 = +5A × 0.2 = +1.0A.

[0047] The calibration proposal for the topology verification domain will be recorded as follows: It is recommended to correct the measured values ​​of electronic transformers T1 down by 2.5A, T2 down by 1.5A, and T3 down by 1.0A.

[0048] The confidence level of the topology verification domain is determined based on the integrity and update frequency of the power grid topology data; the ratio of the number of modeled nodes to the total number of nodes to be modeled in the power grid topology data is evaluated to calculate the power grid topology integrity index; the latest version timestamp of the power grid topology data is obtained, and the topology freshness index, which is the number of days between the latest version and the current time, is calculated; the topology integrity index and the topology freshness index are weighted and averaged to determine the confidence level of the topology verification domain calibration proposal.

[0049] The apparent power calculation value of the electronic transformers in each node and the apparent power measurement value of the merging unit are calculated. The apparent power calculation value is used to characterize the theoretical power restored based on a single measurement. The apparent power measurement value is used to reflect the actual monitoring value of the real-time output power of the electronic transformer, that is, the actual monitoring result of the output capacity at the current moment. The effective voltage value and effective current value of the electronic transformer connected in each node are obtained. The corresponding effective voltage value and effective current value are multiplied to calculate the apparent power calculation value of the electronic transformer at the current moment. Calculate the relative error between the apparent power measurement value and the apparent power calculation value, and filter out nodes whose absolute value of the apparent relative error exceeds a preset error threshold; subtract the apparent power calculation value from the apparent power measurement value obtained at each node to determine the relative error of each node; filter the nodes through the error threshold, and the electronic transformer measurement values ​​of the filtered nodes have offsets.

[0050] It should be explained that the error threshold is set based on the apparent power relative error distribution obtained statistically during historical normal operation periods.

[0051] The second corrected predicted value of the electronic transformer is calculated based on the relative error of each node. The second corrected predicted value includes the second amplitude corrected predicted value and the second phase corrected predicted value. These are then summarized to determine the calibration proposal for the electrical quantity association domain. The proportion of power measurement value of each electronic transformer in the node is calculated, and each proportion is used as an allocation factor to distribute the relative error value. Each allocated relative error value is used as the second corrected original value for each electronic transformer at that node. The second corrected original value is multiplied by the fundamental positive sequence current amplitude of the electronic transformer to determine the second amplitude corrected predicted value. The second corrected original value is then converted by combining the phasor change characteristic factor of the corresponding electronic transformer to obtain the second phase corrected predicted value for each electronic transformer.

[0052] Based on the synchronization accuracy and data refresh rate of the electronic instrument transformer and merging unit, the confidence level of the electrical quantity association domain is determined. The technical specifications of the electronic instrument transformer and merging unit are consulted to obtain their nominal values ​​for time synchronization accuracy, which are then converted into synchronization quality scores ranging from 0 to 1. Simultaneously, the technical specifications of the merging unit are consulted to obtain the nominal values ​​for the data refresh rate of the electronic instrument transformer and merging unit, which are then converted into refresh rate quality scores ranging from 0 to 1. The arithmetic mean of the synchronization quality scores and the refresh rate quality scores is then calculated to determine the confidence level of the electrical quantity association domain.

[0053] The specific process of conducting cross-domain confidence game across different domains is as follows: All possible combinations of the prediction domain, topology verification domain, and electrical quantity association domain are enumerated. For each domain combination, the calibration proposals in each domain combination are weighted and averaged according to their corresponding confidence levels to obtain the coalition value of each domain combination.

[0054] Iterate through all domain combinations that contain a specific domain. All domain combinations containing a specific domain are divided into four categories: combinations consisting of the domain alone, combinations consisting of the domain and other domains of the same category, and complete combinations containing all three categories of domains. For example, when the specific domain is a prediction domain, it is necessary to iterate through the domain combinations consisting of the prediction domain itself, the domain combinations consisting of the prediction domain and the topology verification domain, the domain combinations consisting of the prediction domain and the electrical quantity association domain, and the domain combinations consisting of all domains.

[0055] Calculate the expected average increment of the coalition value before and after adding each domain combination to a specific domain, and determine the contribution value of each domain; calculate the difference in coalition value before and after adding the prediction domain and take the average to obtain the contribution value of the prediction domain. The topology verification domain and the electrical quantity association domain are processed in the same way to obtain the contribution values ​​of the topology verification domain and the electrical quantity association domain. The contribution values ​​of each domain are normalized to determine the dynamic weight allocation strategy; the three values ​​of prediction domain contribution value, topology verification domain contribution value and electrical quantity association domain contribution value are added together to obtain the total contribution value; each contribution value is divided by the total contribution value to obtain the normalized dynamic weight, namely the prediction domain dynamic weight, topology verification domain dynamic weight and electrical quantity association domain dynamic weight. It should be explained that the sum of the dynamic weights of the prediction domain, the dynamic weights of the topology verification domain, and the dynamic weights of the electrical quantity association domain is always 1.

[0056] The predicted values ​​of each domain of the electronic transformer are weighted and summed based on a dynamic weight allocation strategy to obtain the calibration value of each electronic transformer. Electronic transformers whose absolute values ​​of calibration values ​​exceed a preset discrimination threshold are aggregated to generate an execution decision. For each electronic transformer, its drift prediction value is obtained from the drift prediction domain, its first corrected prediction value is obtained from the topology verification domain, and its second corrected prediction value is obtained from the electrical quantity association domain. Based on the weights in the dynamic weight allocation strategy, the drift prediction value, the first corrected prediction value, and the second corrected prediction value of the same electronic transformer are weighted and summed. The result is the calibration value of the electronic transformer, which includes amplitude calibration value and phase calibration value.

[0057] It should be explained that the discrimination threshold is set comprehensively based on the statistical distribution of calibration quantities in historical calibration records and the minimum tolerance requirements of relay protection devices for measurement accuracy; the discrimination threshold can filter out unnecessary calibration operations caused by random fluctuations or minor deviations.

[0058] The calibration execution module is used to correct the sampled values ​​of the electronic instrument transformer based on the calibration decision, and output the corrected sampled values ​​to the relay protection device. The specific process of correcting the sampled values ​​of the electronic instrument transformer based on calibration decisions is as follows: The calibration decision is analyzed to determine the electronic transformer to be calibrated and the calibration quantities. The calibration quantities are then converted into amplitude calibration coefficients and phase calibration offsets. The amplitude calibration coefficients are used to perform linear scaling correction on the current output sampled values ​​of the electronic transformer; the phase calibration offsets are used to apply digital phase-shift filtering to the sampled value sequence. The amplitude calibration quantities are converted to decimal form, and "1" is added to this decimal form of the amplitude calibration quantity to obtain the amplitude calibration coefficient. The phase calibration quantity's angle fraction value is inverted to obtain the phase calibration offset.

[0059] The merging unit performs real-time calibration on each original sampling sequence of the electronic transformer to be calibrated based on the amplitude calibration coefficient and the phase calibration offset. The merging unit multiplies each instantaneous value of the original sampling sequence by the amplitude calibration coefficient to complete the amplitude calibration. The phase calibration offset is adjusted by applying a digital phase shift algorithm based on an FIR filter to the amplitude-calibrated sampling sequence. The new sampling value after calibration is encapsulated into a sampling value message according to the standard format and finally sent to the corresponding relay protection device through the process layer network.

[0060] The strategy learning module is used to evaluate the long-term benefits of smart electronic transformers that have performed calibration decisions, and to dynamically adjust the dynamic weight allocation strategy in cross-domain confidence games based on the obtained evaluation results.

[0061] The specific process for evaluating the long-term benefits of smart electronic instrument transformers that have already undergone calibration decisions is as follows: Acquire relay protection action records for all associated calibrated smart electronic instrument transformers during the evaluation period; relay protection action records include correct actions and incorrect actions. Correct actions indicate valid tripping behavior under the premise that both protection logic and fault location meet the criteria, while incorrect actions include erroneous triggering, missed operation, and failure to operate.

[0062] The accuracy rate of relay protection device operation is obtained based on relay protection operation records; all operation records are statistically analyzed, and the proportion of correct operation to total operation is calculated to form the accuracy rate of relay protection device operation.

[0063] The number of electricity metering disputes caused by excessive measurement error of electronic instrument transformers during the same evaluation period is statistically analyzed to determine the dispute-free rate. The total number of disputes directly caused by excessive measurement error of electronic instrument transformers during the same evaluation period is statistically analyzed from the electricity metering management database. The dispute-free rate is determined by the formula "dispute-free rate = 1 ÷ (number of disputes + 1)".

[0064] The geometric mean of the action accuracy rate and the dispute-free rate is calculated to obtain the long-term benefit assessment value of the dynamic weight allocation strategy. The long-term benefit assessment value is used to represent the actual benefit level generated by the current dynamic weight allocation strategy in the long-term operation from two dimensions: the reliability of protection actions and the accuracy of power metering. The closer the long-term benefit assessment value is to "1", the better the overall performance of the strategy; the closer it is to "0", the worse the effect of the strategy.

[0065] The long-term return assessment value is compared with a preset assessment threshold. If the long-term return assessment value is less than the preset threshold for several consecutive assessment periods, the dynamic weight allocation strategy is updated.

[0066] It should be explained that the update is only triggered when the long-term return assessment value fails to meet the standard for a certain number of consecutive assessment periods. This is to exclude short-term performance decline caused by accidental fluctuations or temporary disturbances in the power grid.

[0067] It should be explained that the preset evaluation threshold is set based on the minimum comprehensive performance requirements of the power grid for the reliability of relay protection operation and the accuracy of power metering.

[0068] The specific process for updating the dynamic weight allocation strategy is as follows: Compare the confidence levels of the prediction domain, topology verification domain, and electrical quantity association domain, and identify the domain with the lowest confidence level as the domain to be adjusted; calculate the relative difference between the confidence level of the domain to be adjusted and the average confidence level to determine the confidence contribution deviation of the domain to be adjusted. The confidence contribution deviation is mapped based on the mapping rules to determine the reduction coefficient of the domain to be adjusted. The mapping rules set the reduction coefficient corresponding to each confidence contribution deviation threshold interval. The division of each confidence contribution deviation threshold interval is based on the normal fluctuation range of the confidence of each domain in historical operation. The larger the confidence contribution deviation interval, the lower the reduction coefficient, so as to achieve the punitive adjustment of the persistently low performance domain. The specific value of the reduction coefficient is determined through simulation test.

[0069] For example, the preset mapping rule is: when the confidence contribution deviation is in the interval [10%, 20%), the reduction factor is 0.9; when the confidence contribution deviation is in the interval [20%, 30%), the reduction factor is 0.8; when the confidence contribution deviation is greater than 30%, the reduction factor is 0.7.

[0070] The reduction factor is loaded into the next round of cross-domain confidence game to reduce the coalition value corresponding to the domain to be adjusted, thereby completing the adjustment of the dynamic weight allocation strategy. In the next round of cross-domain confidence game, when calculating the coalition value of any combination of domains that includes the domain to be adjusted, the calculated original coalition value is multiplied by the reduction factor, and the reduced coalition value is used to participate in the calculation of the contribution value of the domain to be adjusted. Finally, normalization is performed to generate a new dynamic weight allocation strategy.

[0071] In this embodiment, by setting a benchmark construction module, multi-condition composite test signals are applied to the electronic instrument transformers during the initial commissioning stage, and parameter fitting is performed to generate a digital gene vector covering the initial error, temperature drift coefficient, load influence coefficient, and inherent uncertainty at each frequency point. This vector serves as a permanent identifier for the individualized error characteristics of the instrument transformers. By performing time series prediction using historical ambient temperature and load data, and based on the environmental influence factors at future moments, combined with the digital gene vector, quantitative prediction of the future multi-frequency drift error of each instrument transformer is achieved. The drift confidence of the predicted values ​​is evaluated through uncertainty parameters, providing a predictive basis and reference strength for subsequent calibration actions. This solves the problem of passive error correction that traditionally requires calibration to be triggered only when measured values ​​exceed limits. It enables early intervention before actual error fluctuations reach the tolerance, effectively ensuring the response accuracy of relay protection devices in the early stages of faults and improving the overall system's initial stability and operation and maintenance efficiency.

[0072] By constructing a multi-source calibration proposal game mechanism that integrates the prediction domain, topology verification domain, and electrical quantity correlation domain, and generating a dynamic weight allocation strategy based on the confidence level of each domain, the quantification and integration of the contributions of different calibration paths are effectively realized. By statistically analyzing the long-term operational performance indicators of relay protection operation accuracy and electrical metering disputes, the integration strategy is periodically evaluated and dynamically optimized. When operating conditions or main causes of error change, the importance weights of each data domain can be automatically adjusted, thereby ensuring that the final calibration decision always maintains convergence and adaptability. This achieves dynamic coordination, self-feedback weight adjustment, and sustainable calibration capabilities of multi-path decision sources, enhancing the intelligence level and long-term applicability of the electronic transformer calibration system.

[0073] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0074] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only one method, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0075] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

[0076] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A self-calibration system for intelligent electronic instrument transformers for relay protection, characterized in that, include: The benchmark construction module is used to analyze the initial error and drift coefficient of electronic transformers in the early stage of commissioning and generate digital gene vectors. The drift prediction module is used to predict the drift of the electronic transformer based on its historical operating data and digital gene vector, determine the drift prediction value of the electronic transformer in the prediction domain, and determine the drift confidence level based on the inherent uncertainty in the digital gene vector. The game decision module is used to obtain calibration proposals and corresponding confidence levels from the topology verification domain and the electrical quantity association domain, respectively, to conduct cross-domain confidence game for each domain, and to determine the calibration decision based on the determined dynamic weight allocation strategy. The calibration execution module is used to correct the sampled values ​​of the electronic instrument transformer based on the calibration decision, and output the corrected sampled values ​​to the relay protection device. The strategy learning module is used to evaluate the long-term benefits of smart electronic transformers that have performed calibration decisions, and to dynamically adjust the dynamic weight allocation strategy in cross-domain confidence games based on the obtained evaluation results.

2. The intelligent electronic instrument transformer self-calibration system for relay protection according to claim 1, characterized in that, The specific process for generating digital gene vectors is as follows: Set several operating conditions including specific ambient temperature values ​​and specific primary current load levels; apply a composite test signal to the primary side of the electronic transformer under each operating condition. The composite test signal contains standard parameters at each frequency point; synchronously record the synchronous sampling value sequence of the secondary side of the electronic transformer and perform discrete Fourier transform to obtain the output parameters. The standard parameters at each frequency point are compared with the corresponding output parameters to determine the initial error of each electronic transformer at each frequency point; the parameter fitting is performed on the variation law of the initial error at each frequency point with the operating conditions to determine the temperature drift coefficient, load influence coefficient and the inherent uncertainty of the fitting process. By summarizing the initial errors, temperature drift coefficients, load influence coefficients, and inherent uncertainties, a digital gene vector for each electronic transformer is generated.

3. The intelligent electronic transformer self-calibration system for relay protection according to claim 2, characterized in that, The specific process for determining the drift prediction value of the electronic transformer in the prediction domain is as follows: Time series analysis is performed on the historical operating data of each electronic instrument transformer to obtain the predicted values ​​of ambient temperature and load current at specific future moments. Based on the temperature drift coefficient and the predicted load current, the predicted ambient temperature and the predicted load current at each frequency point are multiplied to determine the temperature effect drift component and the load effect drift component at each frequency point. The temperature effect drift component and the load effect drift component are superimposed with their corresponding initial errors to determine the drift prediction value of the electronic transformer in the prediction domain. The drift prediction value includes the amplitude error prediction value and the phase error prediction value.

4. The intelligent electronic instrument transformer self-calibration system for relay protection according to claim 3, characterized in that, The specific process for obtaining the predicted values ​​of ambient temperature and load current at a specific future time is as follows: Time series smoothing decomposition was performed on the environmental temperature series and load rate series of historical operating data to determine the horizontal component, trend component and seasonal component of each series at the current moment. Extrapolate the horizontal and trend components of each sequence at the current moment to obtain the basic predicted values ​​for a specific future moment; Based on the repetitive patterns of the corresponding seasonal components, the predicted values ​​of the basic operating conditions are superimposed and adjusted to obtain the predicted values ​​of the ambient temperature and load current of each electronic transformer at a specific time in the future.

5. The intelligent electronic instrument transformer self-calibration system for relay protection according to claim 4, characterized in that, The specific process for determining the drift confidence level based on the inherent uncertainty in digital gene vectors is as follows: The missing rate and anomaly rate of the ambient temperature series and load rate series in historical operating data are analyzed to determine the basic confidence level of each electronic transformer. Based on the basic confidence level, the inherent uncertainty of the temperature drift coefficient and load influence coefficient is corrected. The uncertainty of the drift prediction value of the corrected temperature drift coefficient and load influence coefficient is quantified to determine the confidence level of the drift prediction value.

6. The intelligent electronic instrument transformer self-calibration system for relay protection according to claim 1, characterized in that, The specific process for obtaining calibration proposals and confidence levels from the topology verification domain and the electrical quantity correlation domain, respectively, is as follows: Acquire the power grid topology data where the electronic instrument transformer is located and the current measurement data of the nodes in the power grid topology; verify the current measurement data using Kirchhoff's Current Law based on the power grid topology data, and obtain the current imbalance residuals of the nodes; Nodes whose absolute value of current imbalance residual exceeds a preset residual threshold are selected, and the proportion of current imbalance residual to each connected electronic transformer in the node is calculated. Based on the proportional distribution of current imbalance residuals, the first corrected prediction value of the electronic transformer in each node is obtained. The first corrected prediction value includes the first amplitude correction prediction value and the first phase correction prediction value. These are then summarized to form a calibration proposal for the topology verification domain. The confidence level of the topology verification domain is determined based on the integrity and update frequency of the power grid topology data. Calculate and obtain the apparent power values ​​of the electronic transformers in each node and the apparent power measurement values ​​of the merging unit; Calculate the relative error between the apparent power measurement value and the apparent power calculation value, and filter out nodes whose absolute value of the apparent relative error exceeds a preset error threshold; The second corrected prediction value of the electronic transformer is calculated based on the relative error of each node. The second corrected prediction value includes the second amplitude correction prediction value and the second phase correction prediction value. These are then summarized to determine the calibration proposal for the electrical quantity correlation domain. The confidence level of the electrical quantity association domain is determined based on the synchronization accuracy of the electronic instrument transformer and the merging unit, as well as the data refresh rate.

7. A self-calibration system for intelligent electronic instrument transformers for relay protection according to claim 5, characterized in that, The specific process of conducting cross-domain confidence game for each domain is as follows: Enumerate all possible combinations of the prediction domain, topology verification domain, and electrical quantity association domain, and calculate the weighted average of the calibration proposals in each domain combination according to their corresponding confidence levels to obtain the coalition value of each domain combination. Iterate through all domain combinations that contain a specific domain, calculate the expected average increase in alliance value before and after the specific domain is added to each domain combination, and determine the contribution value of each domain; normalize the contribution value of each domain and determine the dynamic weight allocation strategy. The predicted values ​​of each domain of the electronic transformer are weighted and summed based on a dynamic weight allocation strategy to obtain the calibration value of each electronic transformer; the electronic transformers whose absolute values ​​of calibration values ​​exceed the preset discrimination threshold are aggregated to generate an execution decision.

8. A self-calibration system for intelligent electronic instrument transformers for relay protection according to claim 7, characterized in that, The specific process of correcting the sampled values ​​of the electronic instrument transformer based on calibration decisions is as follows: The calibration decision is analyzed to determine the electronic transformer to be calibrated and the calibration quantity; the calibration quantity is converted into amplitude calibration coefficient and phase calibration offset, and then sent to the corresponding merging unit. The merging unit performs real-time calibration on each original sampling sequence of the electronic transformer to be calibrated based on the amplitude calibration coefficient and the phase calibration offset, and encapsulates the calibrated sampling values ​​into a sampling value message and outputs it to the relay protection device.

9. A self-calibration system for intelligent electronic instrument transformers for relay protection according to claim 8, characterized in that, The specific process for evaluating the long-term benefits of the smart electronic instrument transformer that has already undergone calibration decisions is as follows: Obtain relay protection action records for all associated calibrated smart electronic instrument transformers during the evaluation period; relay protection action records include correct actions and incorrect actions; obtain the action accuracy rate of the relay protection device based on the relay protection action records; The number of electricity metering disputes caused by out-of-tolerance measurements of electronic instrument transformers during the synchronous evaluation period is counted to determine the dispute-free rate; the geometric mean of the action accuracy rate and the dispute-free rate is calculated to obtain the long-term benefit evaluation value of the dynamic weight allocation strategy; the long-term benefit evaluation value is compared with the preset evaluation threshold, and if the long-term benefit evaluation value is less than the preset threshold for several consecutive evaluation periods, the dynamic weight allocation strategy is updated.

10. A self-calibration system for intelligent electronic instrument transformers for relay protection according to claim 9, characterized in that, The specific process for updating the dynamic weight allocation strategy is as follows: Compare the confidence levels of the prediction domain, topology verification domain, and electrical quantity association domain, and identify the domain with the lowest confidence level as the domain to be adjusted. Calculate the relative difference between the confidence level of the domain to be adjusted and the average confidence level to determine the confidence contribution deviation of the domain to be adjusted; The confidence contribution deviation is mapped based on the mapping rule to determine the reduction coefficient of the domain to be adjusted. The mapping rule sets the reduction coefficient corresponding to each confidence contribution deviation threshold interval. The reduction factor is loaded into the next round of cross-domain confidence game to reduce the alliance value corresponding to the domain to be adjusted, thereby completing the adjustment of the dynamic weight allocation strategy.

Images